PROCEEDINGS of the HUMAN FACTORSSOCIETY34th ANNUAl MEETING-1990

SOCIAL INFLUENCE AND PREFERENCE OF DIRECT-MANIPULATION AND KEYBOARD-COMMAND COMPUTER INTERFACES Michael S. Wogalter and Richard L. Frei Department of Psychology Rensselaer Polytechnic Institute Troy, New York 12180

ABSTRACf Direct-manipulation and command-based computer interfaces have each found their own following among users. This study explores some of the differences between these two groups of computer users. Participants completed a questionnaire that requested their microcomputer usage and ownership, usage and preference of various command methods and pointing devices, the most of their friends use, the microcomputer they would be most willing to purchase next, and their preference for several models of microcomputers. The results showed that participants preferred pointing devices (e.g., mouse) compared to other input methods (e.g., arrow keys) regardless of their prior usage. They tended to use an interface similar to that of their friends' and they reported greater willingness to purchase a computer with an interface similar to the one they most often use. In general, the results suggest that social influence and interface familiarity are important factors in determining which interface people choose to use. Being surrounded by others who use a similar computer interface eases the burden (in terms of effort, time, and expense) of obtaining relevant computer information. An implication of this work is that these variables may hinder approval and acceptance of improved computer interface designs offered by human factors specialists.

INTRODUCfION

Human factors specialists have dedicated considerable and are likely to develop a relationship with these persons. Thus, attention and research to various aspects of computer interface people will tend to affiliate with others, in part, because they have usability. Currently, the marketplace offers two types of interface similar preferences in computers. (Shneiderman, 1987): keyboard-command (e.g., IBM and IBM- clone personal computers) and direct-manipulation (e.g., Apple In addition, Pace and Allison (1990) has recently shown that and Commodore Amiga computers). Keyboard- people tend to value things (e.g., possessions) associated with command interfaces tend to use character-based input and displays their in-group more than objects associated with their out-group. and usually employ keyboard arrow keys to position a curser; Therefore. with greater expertise. users of one interface should direct-manipulation interfaces tend use graphic displays, and increase their liking of the familiar interface and increase their selection of commands using a pointing device (e.g., mouse). disliking of the less familiar computer interface. Thus, the social Both kinds of interface seem to offer advantages and psychological model predicts that people will develop an in- disadvantages. For example, research suggests that experts can group/out-group attitude toward computer interfaces and the issue keyboard commands faster with a keyboard-command people who use them. interface (Karat, McDonald, & Anderson, 1986); whereas, novices learn direct-manipulation interfaces at a faster rate (e.g., So, while both interfaces have certain advantages, people's Brems & Whitten, 1987). The kind of pointing method also choice of which computer to use and to purchase may be due to affects speed and accuracy. Albert (1982) found that in a cursor factors other than those that have been the main concern of human positioning task, faster methods of input (touchscreen, trackball, factors specialists, that of interface design issues such as usability joystick) tended to be the least accurate, while the slowest method and ease-of-use. The purpose of the present report is to describe of input (keyboard) was the most accurate. Nonperformance some evidence for the parallel of these social psychological attributes are also important, such as the user's satisfaction and notions with respect to direct-manipulation and keyboard- preference. However, satisfaction and preference do not always command interface users. concur with performance. For example, Mack and Lang (1989) found that although subjects made significantly more precision Specifically, the present research examines whether users of pointing errors using a stylus and a mouse, they preferred these direct-manipulation and keyboard-command interfaces differ with two input methods over the keyboard-command input method. respect to a number of computer-related factors. Of particular interest is the relationship of prior interface usage (familiarity) and In the last several years, two camps of computer users have others' social influence on choice of microcomputer interface and developed, with each group championing one interface over the purchase intentions. In addition, the present research examines other. Both the direct-manipulation and the keyboard-command whether the direct-manipulation and keyboard-command interface users claim that their interface is better than the other. Emotional users differ in their pattern of usage and preference for computer debate aside, the phenomenon itself appears to be similar to the in- command methods, input devices, time engaged in specific group/out-group bias described by social psychologists. A large activities, and preference for particular microcomputers. body of social psychology research (see Berscheid, 1985) suggests that people like and tend to affiliate with others who have ME1HOD similar likes and dislikes. Conversely, people have greater Participants disliking of others who do not share the same attitudes. This model of social behavior makes several predictions. Because One-hundred thirty-four students participated. Forty-six were computers involve considerable learning time and effort to from Rensselaer Polytechnic Institute (Troy, New York) illld 88 develop expertise, people may seek assistance from others who were from Rice University (Houston, Texas). they like (e.g., friends, coworkers) to facilitate their computer- related decisions (e.g., what hardware and software to buy). To Materials and Procedure balance the relationship, novice users are likely to seek affiliation with persons who previously helped them, and thus, are more The questionnaire was designed and written to avoid showing likely to develop social relationships. As users become more any bias toward a particular kind of interface. Most items were expert on a particular interface, they will then tend to help others,

907 PROCEEDINGS of the HUMAN FACTORS SOCIE1Y 34th ANNUAL MEETING-1990 open-ended in that participants generated (wrote in) the kinds of Space was provided next to the items in the questionnaire for computers that they own and/or use, that they wish to purchase, responses. For items requesting names of computers, participants and that their friends use. Only the last item of the questionnaire were given instructions emphasizing that they try to give both the named particular computer makes and models. The specific make and model. For the items requesting percentages, questionnaire items are shown below: participants were instructed to be sure that the sums totaled 100%. (1) Microcomputer owner: Do you presently own a micro- Virtually all of the participants complied with this request and the data from the few that did not were included "as is" in the computer? analyses. For some items (questions 5, 6, 7, 9, & 13), "other" (2) Time of ownership: If you presently own a was listed as an additional alternative response and participants microcomputer, please list which !dnd(s) of microcomputer(s) were told to describe their answer if they selected this answer. you presently own and how long you have owned each (in "Other" was selected very infrequently and was not counted in the months). analyses. Participants were told that if they were unsure of an (3) Microcomputer use: Please list the microcomputers that answer, that they should give their best guess. you have used over the past six months. Please list them in The questionnaire also requested demographic information an order that corresponds to how much you have used them. (e.g., sex, age, handedness) and other items concerning use of Also, next to each please give your best estimate of how , minicomputers, and mainframe computers. These much total time you have spent using each in total number variables were not found to be significant in any of the analyses of hours for the past six months. and, thus, are not discussed in this report. (4) Microcomputer learned first: What kind of microcomputer did you learn to use fIrst? RESULTS (5) % time in specific activities: In the past 6 months, Categorization of Microcomputers what percentage of your time using microcomputers did you spend for the following activities: (a) accounting, (b) In an earlier pilot study, names of 11 makes and models of computer aided design (CAD), (c) communications, (d) microcomputers were sorted into two piles by 12 Rensselaer desktop publishing, (e) electronic mail, (f) games, (g) Polytechnic Institute students. They were told that in one pile, graphics, (h) music, (i) programming, U) spreadsheets, (k) they should place all those computers that are primarily statistical analysis, and (I) word processing. graphically-based and which usually employ an external pointing device to select commands from menus (direct-manipulation (6) % time using command methods: For the following interface). In the other pile, they should place those kinds of kinds of computer command methods, what percentages of computers that are primarily character-based, and which usually time in the past 6 months did you use to issue commands to employ keyboard arrow keys to position the curser and typed micro-computers: characters to issue commands (keyboard-command interface). a) Typed commands (words and letters followed by the Based on the sorting data, it was found that the Apple RETURN or ENTER key) and Commodore Amiga computers tended to cluster b) Simultaneously-pressed command keys (e.g., CtrI-C, together, and were categorized as direct-manipulation computers. Control-C, CommandoS, etc.) The IBM PCs, PC clones (e.g., AT&T, Compaq, Zenith), and c) Keyboard function keys (e.g., Fl key, page down key, etc.) Apple II computers tended to cluster together, and were d) Choosing commands from menus using the keyboard categorized as keyboard-command computers. When participants e) Choosing commands from menus using a pointing device named specific microcomputers on the questionnaire, the computers were placed into one of the two computer interface (e.g., mouse, trackball). categories. (7) Preference for command methods: How much do you like or prefer the command methods? The same list in Categorization of Interface Users question 6 was given. A 9-point Likert-type scale was given with the following numerical and verbal anchors: (0) Several methods were used to classify participants into the extremely dislike, (1) very much dislike, (2) slightly dislike, two groups according to the type of microcomputer interface they (4) neutral, (5) slightly like, (6) like, (7) very much like, and used. The classification schemes were based on the following (8) extremely like. criteria: the kind of computer they personally own, the kind of computers they use most often, and their preferences. Because (8) % time using pointing devices: In the past 6 months, these criteria were significantly intercorrelated (rs ranged from what percentages of time did you use the following pointing 0.40 to 0.58, ps < .01) and because the results were generally devices: (a) arrow keys, (b) joystick, (c) light pen, (d) mouse, consistent regardless of the categorization method employed, the (e) touch pad, (f) touch screen, and (g) trackball. results of only one method of categorization will be discussed (9) Preference for pointing devices: How much do you here: the kind of computer interface participants used most often. like or prefer the pointing devices? The same set listed in This variable was based on the reported hours of computer use. question 8 and the 9-point scale described in question 8 were Participants indicating that they used direct-manipulation used. computers for more hours than keyboard-command computers were classified as direct-manipulation users. Participants (10) Most willing to purchase: What kind of microcomputer reporting the converse were classified as keyboard-command would you be most willing to purchase? user. A few participants could not be assigned to either interface (1I) Friend owners: What kind(s) of microcomputer) do your because: (1) they reported no use of any microcomputer, or (2) friends and/or coworkers usc most often? If several kinds of they reported the same number of hours of usage for computers of microcomputers are used by them, please order your list with both interfaces. These persons were excluded from most analyses the most frequent first and least frequentlasl. described in this report. (12) Preference for several computer makes and models: Interface User Differences Which of the following microcomputers do you like or prefer. The same 9-point scale described in question 10 was used. The Usage time. The mean total hours using the two micro- following microcomputers were listed: (a) Apple II, (b) Apple computer interfaces by the two interface user groups was Macintosh, (c) Commodore Amiga, (d) IBM PCs, (c) PC examined. A 2 (interface user group) x 2 (computer interface) "clones" (such as, AT&T, Compaq, Tandy,and Zenith PCs). mixed-model analysis of variance (ANOVA) showed a significant

908 PROCEEDINGS of the HUMAN FACTORS SOCIETY 34th ANNUAL MEETlNG-1990 main effect of interface users, F(l, 128) = 3.95, P < .05. In significant effect of interface user group (p > .05), but none was general, keyboard-command users spent more time using expected here since participants were told to balance their microcomputers (152.7 hours) than direct-manipulation interface' responses among the command methods so that they totaled users (100.8 hours). However, there was a significant 100%. However, interface user group did enter into a significant interaction, F(l, 128) = 47.05, p < .0001. Direct-manipulation interaction with command method, F(4, 512) = 18.79, p < .0001. users reported using direct-manipulations machines for more Simple effects analysis showed that the direct-manipulation users hours (M = 85.6) than keyboard-command machines (M = 15.2); selected menus using a pointing device significantly more often whereas, keyboard-command users reported using keyboard- than the keyboard-command users. However, the keyboard- command machines for more hours (M = 132.9) than direct- command users reported using typed commands, keyboard manipulation machines (M = 19.8). There was no reliable main function keys, and keyboard menu selection significantly more effect of computer interface (p > .05). than the direct-manipulation users.

User sample. Most of the direct-manipulation users were from Rice University (83%) and most of the keyboard-command Table 1. Mean Percentage Time using Command Methods by users were from Rensselaer Polytechnic Institute (67%), X1(I, N Interface User Group = 130) = 32.39, p < .0001. The effect of school was examined Direct- by entering it as an independent variable along with interface user Keyboard- group. A series of 2 (interface user group) x 2 (universities) Manipulation Command ANOVAs using all of the dependent variables discussed in this Users Users mean report failed to show a substantial number of significant interactions (5%) between these two variables. This indicates 1YPed that, in general, the pattern of interface user effects described commands 27.54 38.14 32.84 below were consistent regardless of school and that the few significant interactions can probably be attributed to chance. Simultaneously-pressed commandkeys 9.14 13.05 11.09 Fony-five percent of the direct-manipulation users were social Keyboard science and humanities majors, 35% were natural/physical science functionkeys 7.21 15.70 11.45 majors, and 20% were engineering majors. Twenty-six percent of Selectingmenus the keyboard-command users were social science and humanities majors, 19% were natural/physical science majors, and 55% were withkeyboard 4.05 15.05 9.55 engineering majors, X2(2, N = 125) = 15.08, P < .0001. Selectingmenus withpointingdevice 51.72 17.60 34.66 First learned interface and ownership. Seventeen percent of the direct-manipulation users first learned to use a microcomputer on a direct-manipulation machine, and the remaining 83% first Preference. Preferences for the command methods showed a learned on a keyboard-command machine. However, all (100%) pattern of effects that was different from usage rates. The 2 x 5 of keyboard-command users first learned to use a microcomputer ANOVA showed only a significant main effect of command using a keyboard-command machine, X2(1, N = 123) = 7.61, P < method, F(4, 512) = 30.38, P < .0001. Newman-Keuls test .01. Thirty-six percent of the direct-manipulation users own showed that menu selection using a pointing device (M = 6.42) some SOlt of microcomputer; whereas, most of the keyboard- was preferred significantly more than keyboard function keys (M command users own a microcomputer (67%), X1(l, N = 130) = = 5.11). Keyboard function keys were preferred significantly 11.72, P < .001. Thise results are not unexpected since more than the other methods which among themselves did not keyboard-command computers have been on the market for a differ (selecting menus with the keyboard, M = 4.66; typed longer period of time and generally are less costly than direct- commands, M = 4.39; simultaneously-pressed command keys, manipulation computers. M = 4.20). No significant main effect of interface user group, or interaction was noted (ps > .05). Purchase intentions. The direct-manipulation users reponed that they would be more willing to purchase a direct-manipulation Pointing Devices microcomputer (80%) than a keyboard-command microcomputer (20%); whereas, the keyboard-command users were more willing Usage time. Percentage time using several pointing (input) to purchase a keyboard-command microcomputer (67%) than a devices by the two interface user groups was examined. The direct-manipulation microcomputer (33%). X2(1, N = 121) = means can be seen in Table 2. A 2 (interface user group) x 7 25.93, P < .0001. (pointing device) mixed-model ANOVA showed a significant

Social influence. Direct-manipulation users tended to have Table 2. Mean Percentage Time using Pointing Devices by more friends/coworkers that use microcomputers with a direct- Interface User Group manipulation interface (86%) than a keyboard-command interface (14%). However, keyboard-command users tended to have more Direct- Keyboard- friends/coworkers that use microcomputers with a keyboard- Manipulation Command command interface (71 %) than a direct-manipulation interface Users Users mean (29%), X2(1, N = 129) = 42.85, P < .0001.

Command Methods Arrowkeys 15.81 51.42 33.61 Joystick 2.05 4.00 3.02 Usage time. Percentage time using different computer Light pen 1.18 2.84 2.01 command methods by the two interface user groups was examined. The means can be seen in Table 1. A 2 (interface user Mouse 77.83 38.98 58.40 group) x 5 (command modes) mixed-model ANOV A showed a Touchpad 0.38 0.60 0.49 significant main effect of command method, F(4, 512) = 32.48, 0.13 p < .0001. Newman-Keuls comparison test showed that use of Touchscreen 0.08 0.19 keyboard commands and pointing devices did not differ, but both Trackball 2.02 0.52 1.27 were used significantly more (ps < .05) than the other methods which did not differ among themselves. The ANOVA showed no

909 PROCEEDINGS of the HUMAN FACTORS SOCIE1Y 34th ANNUAL MEETING-1990 main effect of pointing device, F(6, 768) = 220.30, P < .0001. spent a greater percentage of their computer time on word Newman-Keuls test showed that the mouse was used significantly processing than the keyboard-command users (p < .05). more than arrow keys. Arrow keys, in turn, were used However, the keyboard-command users spent greater percentages significantly more than the other pointing devices which among of their computer time on computer aided design (CAD), games, themselves did not differ. No significant main effect of interface programming, and spreadsheets than the direct-manipulation users user group was shown, as expected, since these were percentages (ps < .05). (p > .05), but interface user group interacted significantly with pointing device, F(6, 768) = 48.61, P < .0001. Simple effects Additional examination of the activities data was made with analysis showed that direct-manipulation users spent a respect to the number of hours involved. A new variable was significantly greater percentage of their time using a mouse formed based on the product of each participant's reported compared to keyboard-command users. However, keyboard- microcomputer usage hours (during the preceding six months) command users spent a significantly greater percentage of their and the proportions of time allocated to the different activities. time using arrow keys compared to direct-manipulation users (ps The pattern was virtually identical to the percentage of use < .05). The two groups did not differ for the other pointing analysis presented above with one exception: The difference devices (ps > .05). between the two interface user groups for word processing disappeared, showing that both groups spent the same number of Preference. Though the pointing device usage pattern is not hours at this task during the previous six month period. surprising, preference data yielded a different pattern. The 2 x 7 ANOYA showed only a significant main effect of pointing device, Microcomputer Preference F(6, 512) = 48.61, P < .0001. Newman-Keuls test showed that the mouse (M = 6.59) was preferred significantly more than all of Preference for several selected makes and models of the other pointing devices. This was followed distantly by arrow microcomputers by interface user group was examined. The keys (M = 4.73) which was preferred significantly more than all means can be seen in Table 4. A 2 (interface user group) x 5 othcr pointing devices (except touchscreen, M = 4.49). The other (microcomputer) mixed-model ANOYA showed a significant main methods (including touchscreen) did not differ significantly effect of microcomputer, F(4, 512) = 39.82, P < .0001. among themselves (touch pad, M = 4.17; trackball, M = 4.09; Newman-Keuls test showed that the Apple Macintosh was light pen, M = 4.06; and joystick, M = 4.03). There was no preferred most, followed (in respective order) by IBM PCs, PC significant effect of interface user or interaction (ps < .05) clones, Apple II, and Commodore Amiga. All differences were significant (ps < .05) except for the difference between the two Computer-related Activities least preferred computers, Apple II and Commodore Amiga. The ANOYA showed no significant effect of interface user group (p > Percentage time using various computer activities by the two .05), but this variable entered into a significant interaction with interface user groups was examined. The means can be seen in microcomputer, F(4, 512) = 14.10, p < .0001. Simple effects Thble 3. A 2 (interface user group) x 12 (activities) mixed-model analysis showed that the direct-manipulation users preferred the ANOYA showed a significant main effect of specific activities, Apple II and Apple Macintosh computers significantly more than F(ll, 1408) = 107.09, p < .0001. Ncwman-Keuls test showcd the keyboard-command users. However, keyboard-command that microcomputers were used significantly more for word users preferred the IBM PCs and PC clone computers processing than other activities. This was followed by significantly more than the direct-manipulation users. No reliable programming which was significantly greater than the other differcnce bctween groups was found for the Commodore Amiga. activities except for games. Time spent on games was significantly greater than the remaining (lower percentage) activities, followed by spreadsheets which, too, was significantly Table 4. Mean Preference for Selected Microcomputers by Interface greater than the remaining activities. The eight lowest percentage User Group activities did not differ among themselves. The ANOY A also Direct- Keyboard- showed a significant interaction, F(11, 1408) = 18.74, P < .0001. Manipulation Command Simple effects analysis showed that the direct-manipulation users Users Users mean

Table 3. Mean Percentage Time in Specific Activities by Interface AppleII 4.72 3.79 4.26 User Group AppleMacintosh 6.86 5.81 6.34 Direct- Keyboard- CommodoreAmiga 3.99 4.42 4.20 Manipulation Command Users Users mean IBMPCs 5.14 6.14 5.64 PC clones 4.52 5.81 5.17 Accounting 0.23 0.70 0.46 ComputerAidedDesign 0.08 6.21 3.14 Quantity of Interface Usage Communications 1.70 2.65 2.18 The relationship of the quantity of computer interface usage Desktop publishing 1.95 1.37 1.66 and preference was examined. A new variable was created based Electronicmail 1.59 1.65 1.62 on the difference between parti~ipants' reported hours using Games 8.05 16.28 12.16 direct-manipulation and keyboard-command computers in the previous six months. The resulting measure reflected not only Graphics 1.61 3.09 2.35 which interface was used most, but also the magnitude of Music 0.13 0.58 0.35 differential usage. For example, if a participant used a direct- Programming 7.89 17.12 12.50 manipulation computer for 10 hours and a keyboard-command computer for 40 hours, then the score on the new variable would Spreadsheets 1.98 13.77 7.87 be -30. Thus, a high positive score indicates much greater usage Statistical analysis 2.41 0.59 1.49 of direct-manipulation computers and a high negative score indicates greater usage of keyboard-command computers. WordProcessing 68.01 33.42 50.72 Correlations showed that greater usage on one interface was

910 PROCEEDINGS of the HUMAN FACTORS SOCIETY 34th ANNUAL MEETING-1990 associated with greater the willingness to purchase a computer or statistics class), might restrict the kind of computer (and having a similar interface, r (119) = .41, P < .0001, and with software) that may be used for class assignments. However, it is greater likelihood that their friends use a similar interface, r (127) possible that some of the effects noted in this report might have = .36, p < .0001. Participants who spent more time using direct- been stronger if more participants were able to afford (and manipulation computers also had the strongest preference for the purchase) the microcomputer they most wanted. Apple Macintosh computer, r (128) = .43, p < .0001, and those who spent more time using keyboard-command computers also Undoubtedly, the kind of interface that is most desirable had the strongest preference for IBM PCs, r (128) = -.26, p < depends on the task. This was indirectly apparent from the .05, and PC clones, r(128) = -.33, p < .0001. participant's major (e.g., engineering differing from humanities and social sciences), and the results showing that some activities DISCUSSION were used more often by users of one interface than the other.

The evidence collected in the present study parallels the social The present research may have implications for the difficulty psychological notions discussed in the introduction. Computer that human factors specialists often have in implementing users tend to have friends who use the same computer interface as improvements to interfaces. Simply improving an interface may they do. In addition, with increasing use of a particular interface, not be sufficient to gain acceptance. For example, while users participants more strongly preferred it, and more strongly desired may prefer certain aspects of different computer interfaces, people to purchase a computer with a similar interface. Apparently, more may not be willing to make wholesale changes in the computer "expert" users have a stronger in-group/out-group bias with interface they use, even if it reflects a more ergonomically-sound respect to microcomputers. Of course, with the present data, design. The resistance to change might be overcome by cause and effect cannot be separated. At this point, it can not be modifying the currently used interface while still incorporating the determined whether the type of computer people use leads to aspects of the old interface that the user is used to. This is friendships with others using the same interface, or whether apparently happening in the recent evolutions of interface design. friendships with others led to the choice of using one interface Since computers are used for multiple tasks that are often over another (cf. Stafford, 1967). Undoubtedly, effects in both benefitted by different input methods and devices, one way to directions probably occur. integrate users onto a common interface would be to allow many kinds of input mechanism to be connected and integrated A number of interpretations can be offered to explain these coherently so that users can easily switch to the one that best fits social influence relations. For example, being surrounded by the task. However, the major point of the present research is that others who use a similar computer interface eases the burden of factors other than interface design are involved in people's choices accessing new and relevant computer-related information. This of using and owning microcomputers such as prior usage and consideration is particularly relevant given the effort, time, and social influence. expense involved in staying current with today' s rapid computer developments. For example, people may consider the savings of REFERENCES learning time, and the costs of mistakenly purchasing inappropriate software and hardware for their needs. Certainly, it Albert, E. A. (1982). The effect of graphic input devices on is more convenient to ask someone they know for advice and help performance in a cursor positioning task. In Proceedings of the than it is to make these decisions all alone. Human Factor Society of the 26th Annual Meeting (pp. 54-58). Santa Monica, CA: HumanFactors Society. For the most part, the usage and preference data showed a pattern of results that were unexpected given the social Berscheid,E. (1985). Interpersonalattraction. In G. Lindzey and E. psycholological model presented earlier. The results showed that, Aronson (Eds.), Handbook of social psychology, 3rd cd. (pp. 413- in general, participants preferred pointing devices such as the 484). New York: RandomHouse. mouse regardless of their usage rate. So, while usage rates Brems, D. J., & Whitten,W. B. II (1987). Learningand preferencefor showed an interaction indicating that keyboard-command users -basedinterface.In Proceedings of the Human FaclOrs Society employ arrow keys and several command methods relatively of the 31st Annual Meeting (pp. 125-129). Santa Monica, CA: often, the preference data showed no such corresponding effects. HumanFactors Society. These results indicate that the patterns of usage and preferences are dissociatcd, at least with respect to input methods. Moreover, Janis, I. L. (1968). Group identificationunder conditionsof external this finding suggests that keyboard-command users might use a danger.In D. Cartwright and A. Zander (Eds.), Group dynamics: mouse if it were available. Recently, more direct-manipulation Research and theory (pp. 80-90). New York:Harper& Row. features such as the mouse are becoming available in systems that Karat, J., McDonald,J. E., & Anderson,M. (1986). A comparisonof in earlier years were exclusively keyboard-command based. menu selection techniques: touch panel, mouse and keyboard. Some caution is appropriate when interpreting the results of International Journal of Man-Machine Studies, 25, 73-88. this research. The participants were university students who are Pace, J. H., & Allison, S. T. (1990). Positive distinctiveness and probably not representative of all users of microcomputers. intergroup discrimination between intercollegiate athletes and Clearly, some of the results are due to the environmental and nonathletes. Journal of Social Behavior and Personality, 5, 45-60. situational factors that influence the decision of which Shneiderman, B. (1987). Designing the user interface: Strategies for microcomputers to use. Many students have inadequate finances which prevent them from purchasing a microcomputer or effective human-computer interaction. Reading, MA: Addison- constraining their choice to less expensive models. Thus many Wesley. students are limited to those computers that are available the Stafford, J. (1966). Effects of group influences on consumer brand campus facilities and to those owned by friends. In addition, for preferences.Journal of Marketing Research, 3, 68-75. some courses which involve computers (e.g., a computer science

911